Customer storyOracle · AI Agents

Your employees are waiting weeks for the support they need. Our agents deliver the same support in minutes.

Dayos  ·   ·  6 min read
Watch — 10 service requests, under 24 hours eachVideo

How AI agents cleared 10 stuck Oracle service requests for one enterprise customer.

The challenge

When a large enterprise customer saw our IMDA case study, where our platform’s reasoning automated 60% of our own internal Oracle support tickets, they asked a simple question: “Can you do the same for us?”

Their reality was frustratingly common in the enterprise world. They were stuck with a managed service provider (MSP) where end-user service requests sat in the queue for an average of 21 days.

To test our platform, they handed us 10 real, currently stuck Oracle service requests. These weren’t simple password resets; they were complex data extracts, workflow routing challenges, and production report modifications.

Here is exactly what they asked for, and how our platform’s AI agents handled the heavy lifting.

10 real requests, handled by AI

The ask → the execution
01Receivables

The ad-hoc data extract

The ask

A plain-language request for a specific receivables report containing all necessary fields.

Execution

No forms, no complex report builders, and no waiting on a developer. The platform instantly figured out where the data lived, gathered it, and returned a clean, ready-to-use file with 1,800 lines. With one additional prompt, that raw data was turned into a polished, formatted report ready to share.

02Procurement

Surgical updates to production reports

The ask

Add missing project details behind every purchase order line on an existing PO report.

Execution

The platform opened the live report, identified the necessary data connections, and injected the missing details without breaking anything already in production. A real change to a live report, executed from a single plain-language request.

03Receivables

Massive customer report cleanup

The ask

Clean up a heavily outdated customer report.

Execution

This was a heavy lift requiring dozens of simultaneous changes: adding missing fields, renaming columns to match current standards, and dropping deprecated data. The platform reworked the entire report in a single pass and handed back a working version. This level of cleanup normally ties up a developer for days.

04Receivables

Dynamic filtering for finance

The ask

Turn a static, printed list into a dynamic report the finance team can actually work with.

Execution

The platform added new breakdowns and every filter the team requested (by account, by type, by cost center, etc.). Now, the finance team can pull the exact view they need instead of scrolling through a fixed list.

05Cash Management

Automated bank reconciliations

The ask

Automate manual bank reconciliations, where transactions are matched against statements line by line.

Execution

Matching gets messy: sometimes one payment shows up as many, and sometimes many show up as one. The platform automated the reporting across all these scenarios and built alerts to flag exceptions. Now, the team only steps in when something truly doesn’t line up.

06Payables

Intelligent AP approval routing

The ask

Ensure every invoice automatically reaches the correct person for approval.

Execution

The platform digested the company’s internal approval rules and built the corresponding routing logic into the system. Every invoice now finds the correct approver automatically: no more manual chasing.

07Scheduler

Human-in-the-loop for security

The ask

Pull a list of every scheduled process running in the system.

Execution

This request hit a roadblock: some of the necessary data was locked down by Oracle for valid security reasons. Instead of guessing or failing silently, the agent did exactly the right thing. It raised a ticket and pulled a human engineer into the loop to confirm details and grant access. Once sorted, the agent picked the job right back up and delivered the full list.

Human-in-the-loop when it matters; full automation the rest of the time.

08Security

User access reviews

The ask

Generate a clean listing of who is in the system, when they last logged in, and the jobs they have scheduled.

Execution

The platform pulled all of this data into a single, clean listing that is now ready to run at any time.

09Security

User access reviews

The ask

Generate a clean listing of who is in the system, when they last logged in, and the jobs they have scheduled.

Execution

The platform pulled all of this data into a single, clean listing that is now ready to run at any time.

10Payables

Intelligent AP approval routing

The ask

Ensure every invoice automatically reaches the correct person for approval.

Execution

The platform digested the company’s internal approval rules and built the corresponding routing logic into the system. Every invoice now finds the correct approver automatically, no more manual chasing.

The outcome

Faster, cheaper, better

The results across these 10 requests were transformative.

<24HAverage turnaround

Every single ticket was automated and resolved in under 24 hours. Why the dramatic drop from 21 days? Because with our platform, no ticket ever waits in a queue for an engineer or admin to free up.

<$10Per ticket

The compute cost to resolve these complex requests was under $10 each, using roughly 1.97M tokens.

2 WKSTo deploy & prove

In just 14 days, the customer had their own tailored case studies proving that 60% of their monthly tickets could be resolved automatically.

The takeaway

For this enterprise customer, the days of waiting 21 days or longer to get the configurations, reports, and analysis needed to run their business are over.

AI agents are no longer a future promise. They are actively doing the heavy lifting in enterprise systems today.